1. F ACE PERCEPTION
1.1 B RAIN STRUCTURES INVOLVED IN FACE PERCEPTION
The superior temporal sulcus (STS) in the macaque brain contains neurons that respond specifically to faces using single-unit recordings (Gross, Roche-Miranda, &
Bender, 1972; Desimone, 1991; Perrett et al., 1991). This evidence has led researchers interested in face perception to question whether a region in the human brain is specialised in this processing as well. Besides the fact that faces seem to be preferentially processed by the individuals (Yin, 1969; Bruce, Doyle, Dench, & Burton, 1991), some neuropsychological conditions in brain damaged patients with specific face processing dysfunctions have also led to the investigation of brain areas linked to face perception (Damasio, Tranel, & Damasio, 1990; Behrmann, Winocur, & Moscovitch, 1992).
Bruce and Young (1986) have proposed that two parallel processing routes are responsible for facial expressions and identity processing. These routes would work independently such that expression can be treated without identity processing (Bauer, 1984; Breen, Caine, & Coltheart, 2000). Support for this model comes from prosopagnosic patients who are still able to recognise expressions (e.g. Damasio, Damasio, & Van Hoesen, 1982; Damasio et al., 1990) and other patients with intact identity recognition but who are unable to process specific facial expressions as for example disgust following insula damage (Calder, Keane, Manes, Antoun, & Young, 2000) or fear following amygdala damage (Adolphs et al., 1995). This model has been questioned by some behavioural data showing that expression evaluation can be affected by familiarity (Schweinberger & Soukup, 1998) and that learning of new face identities is facilitated by expression (Baudouin, Gilibert, Sansone, & Tiberghien, 2000;
Sansone & Tiberghien, 1994). However, face perception seems to activate specialised brain regions that were investigated using functional magnetic resonance imaging (fMRI), which allows the investigation of brain structures involved in the processing of
stimuli with a high spatial resolution. Several brain regions have been highlighted during the processing of faces as compared to other types of stimuli in studies using numerous experimental paradigms. In particular, more important activation in the ventral occipitotemporal cortex (especially the lateral fusiform gyrus) and the inferior occipital gyrus are observed during the processing of faces. The activation of these regions does not appear in isolation and they have often been reported to collaborate in order to categorise and bring to consciousness the concept of a face.
The model proposed by Bruce and Young (1986) was elaborated following observations made in behavioural experiments. Based on fMRI, PET and ERP studies, Haxby, Hoffman and Gobbini (2000) proposed an influential model where a comprehensive representation of human faces is extracted through a joint activation of frontal, temporal and occipital cortices. According to this model, a core system in the visual cortex processes invariant and changeable aspects of faces. This system interacts with an extended system where other brain regions specialised in different aspects of face perception allow the completion of visual analysis. The extended system integrates features related to social communication such as attentional processes, speech perception, emotion and identity-related information (see Fig. 1). The integration of all these aspects would allow face identification.
The organisation of the visual system has been categorised into two distinct pathways emerging from the occipital cortex. The “what” pathway goes along the ventral stream and is responsible for object recognition; the “where” pathway goes through the dorsal stream and reflects processing of spatial vision. This categorisation
Figure 1: The distributed neural system for face perception. According to this model, two systems interact during face perception. The core system refers to visual anlysis of the face
raised the question of whether a specialised region for faces is present along the ventral vision pathway. Haxby et al. (1994) have highlighted that the ventral pathway processes several object categories, whose neural correlates are represented in distinct regions.
Kanwisher, McDermott and Chun (1997) showed activation of the lateral fusiform gyrus to faces as compared to different types of other stimuli, including scrambled faces, houses and human hands. This study showed that activation of this region is not linked to visual attention, stimulus classification or processing of human forms, highlighting its specificity in the processing of faces. Consequently, the lateral fusiform gyrus has been labelled “the fusiform face area”. At the same time, another study used continuously changing montage in which faces, common objects and nonobjects were displayed and showed bilateral activation of the fusiform gyrus when contrasting faces to nonobject processing. Moreover, when contrasting faces to object processing, only the right fusiform gyrus showed significant activation (McCarthy, Puce, Gore, &
Allison 1997), suggesting the specificity of the right hemisphere in face perception.
The inferior fusiform gyrus (IOG), also known as the occipital face area (OFA), in collaboration with the fusiform gyrus and the superior temporal sulcus, seems to be a central neural substrate in the processing of faces. Indeed, the IOG is more active during the processing of faces than other types of stimuli (Pitcher, Walsh, & Duchaine, 2011).
In support of this observation, lesion studies showed that damage in this brain region leads to a deficit in face recognition (prosopagnosia; Bouvier, & Engel, 2006).
However, it is not yet established whether this region is involved in early stages of face processing (which is claimed by some authors based on the observation that IOG is the most posterior region related to face processing (e.g. Pitcher et al., 2011) or in later stages shaping identity recognition (Rossion, Dricot, Goebel, & Busigny, 2011).
Intracranial studies on epileptic patients directly implanted in this brain region recorded event-related potentials during the processing of faces and showed a specific activation for faces in the IOG in the time range generally observed during the processing of these stimuli (~170 ms) (Allison, Puce, Spencer, & McCarthy, 1999; Rosburg et al., 2010), suggesting that this region plays a crucial role in early stages of processing.
Other brain regions have also been observed during the processing of face perception, such as the amygdala (Garvert, Friston, Dolan, & Garrido, 2014), implied in early stages of face perception and the anterior temporal lobe (Von Der Heide, Skipper,
& Olson, 2013) during face identification and memory. Neurons in the orbitofrontal cortex (OFC) who respond selectively to faces have also been reported in the macaque brain (Rolls, Critchley, Browning, & Inoue, 2006). It is suggested that this network is activated differentially depending on task demands (Vuilleumier & Pourtois, 2007), but the way these different regions act to form the concept of a face is still an open question.
Specifically, the timing of activation of these different regions will now be investigated.